J Insect Conserv (2013) 17:183–194 DOI 10.1007/s10841-012-9496-x

ORIGINAL PAPER

A first step towards successful conservation: understanding local oviposition site selection of an imperiled butterfly, mardon skipper

Erica H. Henry • Cheryl B. Schultz

Received: 14 November 2011 / Accepted: 4 May 2012 / Published online: 23 May 2012 Ó Springer Science+Business Media B.V. 2012

Abstract Lack of basic biological information is a key importance of understanding the local habitat use of a rare limiting factor in conservation of at-risk butterflies. In the species where restoration activities occur and increase our Puget prairies of Washington State little is known about the ability to target habitat management where it is most needed habitat requirements of mardon skipper (Polites mardon, for the persistence of the species. federal candidate, WA endangered). We investigated ovi- position site selection and used our results to assess ovipo- Keywords Butterfly Endangered species Habitat sition habitat quality at a restored site with reintroduction requirements Hesperiinae Prairie management potential. During the 2009 flight season we marked eighty- Restoration eight eggs and sampled vegetation at oviposition and ran- dom locations, measuring habitat variables with respect to the oviposition plant, vegetation structure, and vegetation Introduction cover. Eighty-six of the eighty-eight eggs were laid on Festuca roemeri, a native, perennial bunchgrass. Discrimi- Habitat restoration and reintroduction strategies are nant function analysis revealed selection of oviposition sites essential elements of recovery plans for rare and endan- based on habitat structure; females laid eggs in small F. gered species. These efforts often fail, however, because roemeri tufts in sparsely vegetated areas of the prairie. These the ecology of focal species is unknown (Griffith et al. results are contrary to results from a previous study in the 1989; Miller and Hobbs 2007; Morrison 2009; Settele and Cascade Mountains of WA where females are generalists Kuhn 2009). In the absence of detailed knowledge of a and selected densely vegetated areas, suggesting that the species’ ecology, land managers are limited in what they species has geographically specific habitat requirements. To can do to conserve and restore rare species populations. assess oviposition habitat at a potential reintroduction site Often, ecosystems known to contain target species are we measured the six variables most important for oviposi- protected. This strategy can fail, however, when protected tion at the occupied site and a proposed reintroduction site. areas are too small and/or isolated to sustain populations or Results revealed differences in habitat quality between do not maintain crucial habitat elements for target organ- locations and suggest a need for further habitat manage- isms (Barinaga 1990; Hanski 1991; New et al. 1995; ment at the reintroduction site. Our results highlight the Newmark 1996; Panzer and Schwartz 1998; Wenzel et al. 2006; Settele and Kuhn 2009). This is particularly true for animals with specialized habitat requirements such as & E. H. Henry C. B. Schultz ( ) butterflies (Thomas 1984). Washington State University, 14204 NE Salmon Creek Ave, Vancouver, WA 98686, USA The needs of each butterfly life stage (egg, larva, pupa, e-mail: [email protected] adult) are distinct and often occur at different scales (Dennis 2010). The resource based view of butterfly habitat Present Address: defines suitable habitat as an area where adult and larval E. H. Henry Department of Biology, North Carolina State University, resources co-occur within the exploratory range of indi- P.O. Box 7617, Raleigh, NC 27695-7617, USA vidual butterflies (Dennis et al. 2003, 2006; Shreeve et al. 123 184 J Insect Conserv (2013) 17:183–194

2004; Dennis 2010). A suitable habitat patch therefore characteristics that influence oviposition site selection, and must contain the full complement of resources, and opti- (3) use this information to develop a method to evaluate the mally supplemental resources to buffer for stochastic suitability of habitat restored for the mardon skipper. events (Dennis 2010). The loss of the habitat important for any critical resource such as adult food sources, basking locations, and specific microclimatic conditions for larvae, Methods reduces fecundity and/or survivorship (Boggs and Ross 1993; Nylin and Gotthard 1998; Karlsson and Wiklund Study species 2005). In addition to local resources, butterfly populations rely on appropriate landscape conditions such as patch size The mardon skipper is a member of the grass skipper (Crone and Schultz 2003), and for some species the proper subfamily Hesperiinae of which at least 35 species are at- spatial arrangement of habitat patches is vital to maintain risk worldwide (Beyer and Schultz 2010). Mardon skippers metapopulations (Crone and Schultz 2003; Hanski 2003). are endemic to the Pacific Northwest of the United States The specific nature of butterfly habitat requirements has and occur in four disjunct locations: the Puget prairies, led to the use of single species and single population montane meadows in both the Cascade Mountains of conservation approaches (New et al. 1995; Dennis et al. southern Washington and the Siskiyou Mountains of 2008). The most successful of these endeavors come from southern Oregon, and serpentine grasslands in northern systems where detailed knowledge of the ecology of target California (Fig. 1). The Siskiyou Mountain population has species exists, specifically the preferred larval habitat and been described as the distinct sub-species P. mardon optimum larval resource (Thomas et al. 2011), and is klamathensis (Mattoon et al. 1998). The taxonomic status incorporated into conservation strategies (Schultz and of the other three populations has not been evaluated; given Crone 2008; Thomas 1983; Warren 1991; Thomas et al. the highly disjunct nature of the species’ distribution each 2009; Longcore et al. 2010). Within the relatively well- regional population may be taxonomically unique. studied group of temperate butterflies and skippers critical Mardon skipper populations range-wide are isolated and information gaps still exist, particularly for rare species small. The largest populations support 1,000–4,000 but- (New et al. 1995; Garcia-Barros and Fartmann 2009). terflies, but most range from 50–200 individuals (Potter The mardon skipper (Polites mardon, Washington State et al. 1999; Beyer and Black 2006; Jepsen et al. 2008). endangered) is a candidate for listing under the United Mardon skippers are univoltine with adults in the Puget States Endangered Species Act whose survival depends on prairies flying from mid-May to mid-June. Females drop habitat restoration (USFWS 2011). Mardon skipper popu- eggs into the grass singly without affixing them to a par- lations in the glacial outwash prairies of the Puget lowland, ticular location. The eggs hatch 7–10 days later and larvae Washington, USA (hereafter referred to as Puget prairies) feed until late fall when they diapause in dried grass have declined precipitously in recent decades (Potter et al. shelters at the base of the grass (Henry unpublished data). 1999). Because of these declines, prairie restoration in the In spring, larvae feed again before pupating in April region often aims to improve mardon skipper habitat. (Henry unpublished data). However, restoration planning is limited because little is known about the butterfly (but see Beyer and Schultz Study sites 2010). Important nectar species have been identified (Hays et al. 2000), but the host plant, life history, and specific Lack of historic disturbance and subsequent succession of habitat requirements of the butterfly in the Puget prairies grasslands have dramatically reduced mardon skipper habi- remain unknown. tat throughout its range (Noss et al. 1995; Crawford and Hall Mardon skippers are confined to two isolated Puget 1997). Additionally, development and conversion to agri- prairies. As a result, reintroduction has been proposed to culture have shrunk the Puget prairies to 3 % of their pre- return mardon skippers to previously occupied sites where vious distribution (Crawford and Hall 1997). Amerindians habitat restoration has occurred (A. Potter, personal com- historically maintained these prairies by setting frequent, munication). The World Conservation Union guidelines for low intensity fires (Storm and Shebitz 2006). In the absence reintroductions state that habitat quality at target sites must of frequent disturbance, remaining prairie habitat has been be evaluated, and reintroduction should occur only where degraded by the invasion of non-native species such as high quality habitat exists (IUCN 1998). It is impossible, Scotch broom (Cytisus scoparius) and tall oatgrass (Arr- however, to restore habitat or assess habitat quality without henatherum elatius). These species greatly alter the vege- knowledge of the target species’ requirements. In this tation structure and outcompete native plants (Dennehy et al. paper, we aim to (1) identify hostplants used by mardon 2011). To combat invasive species, prairie management in skippers in the Puget prairies, (2) identify habitat the region involves a combination of techniques including 123 J Insect Conserv (2013) 17:183–194 185

Fig. 1 Mardon skipper range. West coast distribution (left) and detail of Puget lowland sites (right)

prescribed burning, mowing, native species planting and Both Scatter Creek and Mima Mounds are part of a invasive species removal through herbicide application and Prairie Quality Monitoring project (Olson 2010). This hand pulling (Dennehy et al. 2011). In the 1980’s mardon project involves mapping vegetation across the entire site skippers occurred in at least eight Puget prairies and now at a 25 m 9 25 m resolution. Roemer’s fescue (Festuca persist at only two, Scatter Creek Wildlife Area and a few roemeri, a native perennial bunchgrass), tall oatgrass, and locations on the edge of the Artillery Impact Area at Joint scotch broom are three of the species for which percent Base Lewis-McChord (Fig. 1) (Potter et al. 1999). Because cover is estimated within each grid cell. The abundance of of active military training, the Artillery Impact Area was not a variety of forb species, including early blue violet (Viola included in this study. adunca), is also measured (Olson 2010). WDFW collected Scatter Creek Wildlife Area (46°5001400 N, 122°5904200 these data at Mima Mounds in the summer of 2008 and at W) includes 250 ha of glacial outwash prairie, and is Scatter Creek in the summer of 2009 (Olson 2010). managed by the Washington Department of Fish and Wildlife (WDFW). In 2009 the north (120 ha of prairie) Oviposition habitat selection and south unit (130 ha of prairie) of Scatter Creek sup- ported populations of 100–300 and approximately 1,000 Oviposition surveys mardon skippers respectively (G. Olson, unpublished data). Prairie quality and species composition varies dramatically We coordinated our oviposition surveys with WDFW across the wildlife area (Olson 2010). biologists conducting mardon skipper distribution and Mima Mounds (46°5304000 N, 123°301200 W; Fig. 1)isa abundance surveys (Potter and Olson 2009) during the Washington Department of Natural Resources Natural Area 2009 flight season (May 15–June 15). WDFW surveyors Preserve located 8 km northwest of Scatter Creek with walked east-west transects spaced 100 m apart and dropped 160 ha of prairie. Mardon skippers were last recorded at a flag at all mardon skipper detection locations (Fig. 2). Mima Mounds in 1994 (Potter et al 1999). State biologists After each of their three surveys, WDFW sent us the have proposed that a series of large (20–80 ha) prescribed locations of all flagged mardon skipper detections. We then burns in the early 1990s was the primary driver of their searched for females by starting at a randomly selected extirpation, although other factors including nectar limita- detection location and haphazardly searching the sur- tion, invasive species, and/or climate may have contributed rounding hectare for 20 min or until the first female was as well (D. Wilderman, personal communication). located. If no female was detected within 20 min the

123 186 J Insect Conserv (2013) 17:183–194

Fig. 2 Mardon skipper detections and ovipositions. Locations of locations (white circles) in 2009 at the north and south units of Scatter Washington Department of Fish and Wildlife (WDFW) mardon Creek Wildlife Area, WA. Black lines are transects walked by skipper survey detections (black circles) and our observed oviposition WDFW surveyors observer proceeded to the next random detection location. sampled vegetation at oviposition locations and a paired Once located, the observer followed the female until she random location 30–45 m away in a random direction. We oviposited, was lost, or 30 min had passed. If a female was chose this distance to capture the variability within the lost or exhibited no oviposition behavior in 30 min, the landscape while remaining within the exploratory range of observer continued to search for additional females at the female butterflies. During pilot observations in 2008 we same location. Once the observer saw oviposition, s/he watched mardon skippers routinely fly up to 50 m (Henry, marked the egg location by placing a flagged wooden unpublished data). We centered meter-square plots on the skewer as close to the egg as possible without disturbing it, oviposition plant and the plant of the same species closest identified the selected grass species, and recorded the GPS to the random location. Within each plot we measured coordinates. During observations of female mardon skip- habitat variables with respect to the oviposition plant, the pers in 2008 we witnessed no ovipositions before 11:00 h; structure of the vegetation, and vegetation cover, all of instead, female butterflies spent the morning hours nec- which influenced mardon skipper oviposition site selection taring (Henry, unpublished data). Therefore, we conducted in the Washington Cascades (Beyer and Schultz 2010). oviposition surveys between 11:00–16:00 h on all days Oviposition plant variables included the footprint (maxi- during the flight season when mardon skippers were mum width 9 length), and height of the tallest leaf in the observed flying. grass tuft (multiplied to get an index of tuft size), number of culms (flowering stalks), maximum culm height, percent Vegetation sampling dead leaves (category 1 = 0–25 % dead, 2 = 26–50 % dead, 3 = 51–100 % dead), distance to nearest plant of the To effectively assess the vegetation community and same species as selected oviposition plant, and ground- structure experienced by mardon skipper females, we cover (ie. thatch, moss) depth immediately adjacent to base sampled vegetation within 2 weeks of oviposition. We of the oviposition plant. Vegetation structure attributes

123 J Insect Conserv (2013) 17:183–194 187 included tallest vegetation species and height, distance to randomly selected one of each pair to eliminate (McGarigal nearest tree, number of individual fescue tufts, number of et al. 2000). Only two pairs of our measured variables were nectar flowers, and vertical vegetation density. We mea- highly correlated: moss cover and thatch cover (0.82), and sured vertical vegetation density with one observer holding maximum culm height and number of culms (0.81). Moss a meter stick parallel to the ground at 20 cm off the ground and thatch are both prevalent prairie ground covers and, as along the north edge of the plot, and a second observer the negative correlation suggests, where one is present, the lying on the ground on the south edge of the plot estimating other is absent; maximum culm height and the number of the % of the meter stick obscured by vegetation. culms are both measures of plant vigor. The resulting Because visual percent cover estimates can vary greatly twenty-two variables were included in the DFA (Table 1). between observers (Sykes et al. 1983), we used the point Only samples with complete data sets were entered in the intercept method (Goodall 1952) to collect vegetation analysis, which left us with 74 oviposition locations and cover data. Based on power analysis of data collected in 82 random locations. Prior to analysis we log10(x ? 1) 2008 we used 33 pins/m2 to estimate vegetation cover transformed continuous data and arcsine square-root (Henry 2010). Our pin frame contained eleven pins along transformed percent cover data; all variables had homo- a meter, one every 10 cm starting at zero. We placed the scedastic variances and we subjectively assessed the vari- frame at three randomly selected locations within the ance among groups with the ordination plot (Quinn and quadrat (either at 10, 30, 50, 70, or 90 cm along the Keough 2002). To determine which variables contributed perimeter) and recorded every species that contacted each most to the separation of groups along the canonical axis pin as well as the ground cover at the base of the pin. we used total structure coefficients, which measure corre- lations between individual variables and the canonical Statistical analysis function, and interpreted variables with the highest abso- lute value coefficients to be most important. We conducted We ran discriminant function analysis (DFA) to determine DFA in SAS 9.2 statistical package using the CANDISC if oviposition and random plots formed independent groups procedure. (Williams 1983). Before running DFA we identified highly In addition to identifying differentiating factors between correlated variables (Pearson correlation [ |0.7|) and oviposition and random locations, we used one-way ANOVAs

Table 1 Discriminant function Variables (units) Total structure Oviposition Random Fpvalue analysis variables and results coefficient locations locations

Tuft size (cm3)1 0.733 6,873 – 7,017 21,136 – 21,167 55.23 <0.001 Number of tufts (count)2 20.662 11.0 – 4.4 6.5 – 3.9 47.99 <0.001 Percent dead (category 1–3)1 0.610 1.4 – 0.6 2.049 – 0.780 32.54 <0.001 Vertical vegetation density (%)2 0.606 14.9 – 19.5 36.3 – 30.1 31.65 <0.001 Tall oatgrass cover (%)3 0.488 2.8 – 5.3 10.3 – 13.0 21.25 <0.001 Moss cover (%)3 20.474 56.1 – 23.0 39.0 – 24.8 19.79 <0.001 Distance to nearest Fescue (cm)1 0.442 17.7 – 8.3 31.1 – 33.4 11.3 0.001 Scotch broom cover (%)3 0.356 2.9 ± 5.9 8.9 ± 14.8 10.61 0.001 Fescue cover (%)3 -0.343 51.6 ± 20.1 39.9 ± 24.4 9.57 0.002 Thatch depth @ ovp plant (cm)1 0.328 3.4 ± 1.9 4.4 ± 2.2 10.6 0.001 Carex sp. cover (%)3 0.316 25.3 ± 16.3 34.8 ± 21.1 8.19 0.005 Variables included in the DFA. Height of tallest vegetation (cm)2 0.288 63.1 ± 15.5 71.2 ± 18.7 8.61 0.004 Variables are ranked by the Total vegetation cover (%)3 0.272 92.9 ± 12.4 97.0 ± 2.9 6.02 0.014 absolute value of their total 1 Number of culms (count) 0.254 1.6 ± 3.4 3.3 ± 6.6 3.74 0.055 structure coefficients. Top 3 discriminating variables are Hypochaeris sp. cover (%) -0.244 6.2 ± 6.5 4.0 ± 4.9 4.25 0.041 bold. p values are significant at Bare ground cover (%)3 0.139 6.6 ± 10.7 7.9 ± 11.1 0.6 0.439 p \ 0.002. Superscripts refer Dry vegetation cover (%)3 0.130 60.6 ± 23.9 65.1 ± 23.6 1.58 0.210 to the type of variable: Nectar flowers (count)2 -0.098 1.5 ± 5.0 0.7 ± 1.3 1.17 0.282 1 oviposition plant, 2 vegetation 2 structure, 3 vegetation cover. Distance to nearest tree (m) -0.082 39.8 ± 30.8 34.8 ± 26.9 1.16 0.283 Variable values are Danthonia sp. cover (%)3 -0.063 4.8 ± 6.2 4.2 ± 5.4 0.33 0.564 mean ± 1 SD. Corresponds Agrostis sp. cover (%)3 0.023 10.4 ± 12.6 11.1 ± 12.7 0.13 0.721 with Fig. 4

123 188 J Insect Conserv (2013) 17:183–194 to examine differences in individual variables between ovi- criteria in which to sample fine scale habitat elements based position and random locations. After transformations, all on recommendations from the Program Ecologist at Mima variables met the assumption of homoscedasticity and we ran Mounds (Fig. 3) (D. Wilderman, personal communication). ANOVAs regardless of normality (Scheffe 1959)(a = 0.002, During the 2010 mardon skipper flight season we measured Bonferroni multiple comparison). variables from the 2009 oviposition versus random DFA analysis with total structure coefficients [ |0.4| except for Developing an oviposition habitat quality assessment tall oatgrass cover which was excluded a priori through the method aforementioned GIS analysis (hereafter referred to as the top discriminating variables). We chose the 0.4 cutoff Habitat assessment based on results from similar analyses of mardon skipper oviposition and random locations performed by Beyer and The effects of habitat management in Puget prairies are Schultz (2010). At all but one of the nine meadows they often measured by comparing native and/or invasive spe- surveyed in the Washington Cascades the top 5 discrimi- cies abundance before and after treatment (Stanley et al. nating variables had structure coefficients [ |0.4| (Table 3 2011). We were interested in developing an easy to in Beyer and Schultz 2010). In our analysis, Scotch broom implement method for use by land managers to assess how and fescue cover are the two variables immediately below management activities impact mardon skipper oviposition the 0.4 cut-off (Table 1) both of which we accounted for habitat quality. Washington State biologists have proposed a priori in the GIS analysis. We measured the top dis- Mima Mounds as a candidate site for mardon skipper criminating variables in eight, meter-square plots, ran- reintroduction (A. Potter, personal communication) so we domly placed along each of five 50 m transects. All chose to collect data there for use in developing our transects were contained within a half-hectare polygon method. Our method involved a two-step approach. We (50 m 9 100 m). Each plot was centered on the potential first used vegetation data from the Prairie Quality Moni- oviposition plant (of the oviposition species observed in toring dataset to find suitable locations within Mima 2009) nearest to each randomly selected transect meter Mounds for potential reintroduction. We conducted GIS mark. In 2010 we also sampled vegetation at the south unit analysis (ArcGIS 9.3) to locate areas of Mima Mounds with of Scatter Creek using the same method in two half-hectare moderate fescue cover (potential host plant), high violet polygons that encompassed as many 2009 oviposition abundance (important nectar source), and no scotch broom locations as possible Fig. 3. This sampling allowed us to or tall oatgrass cover to limit the need for additional weed compare the habitat quality in the locations sampled at control efforts in potential reintroduction sites. We then Mima Mounds to that of an area at Scatter Creek that chose two locations that contained areas that met these supports a relatively high density of mardon skippers.

Fig. 3 Polygons sampled for habitat assessment analysis. Locations of ‘ ha polygons sampled at Mima Mounds and Scatter Creek (black rectangles). Prairie habitat at both locations is outlined in grey. White boxes on Mima Mounds map are results from GIS calculation to locate areas with moderate fescue cover, high violet abundance, and no scotch broom or tall oatgrass cover, individual squares are 25 m 9 25 m. 2009 oviposition locations at Scatter Creek are indicated with white circles

123 J Insect Conserv (2013) 17:183–194 189

yellow hairgrass (Aira praecox) a non-native annual grass. Roemer’s fescue cover accounts for only fifty percent of the total grass cover in all plots sampled suggesting female butterflies are actively selecting to oviposit on fescue over other available grasses.

Vegetation sampling

Discriminant function analysis revealed that oviposition and random locations were different from each other (MANOVA: Wilks’ Lambda = 0.499, F = 6.01, df = 22, p \ 0.0001; r2 (proportion of total variation explained by function) = 0.50; Fig. 4). The variables that contributed most to discrimination between groups were related to the Fig. 4 Discriminant function analysis results. Oviposition locations structure of both the selected fescue tufts and vegetation are grey, random locations are hatched. Oviposition and random within the plot. Females laid eggs in small, relatively green locations form two different groups along the canonical axis. Percent fescue tufts. In the square-meter surrounding oviposition 2 of canonical variation explained by group differences (r ) = 0.50. locations there was little tall oatgrass cover and vertical Variables corresponding to the canonical axis are given in Table 1 vegetation density, and moss is the predominant ground cover (Table 1). All of these variables differ between oviposition and random plots (p B 0.001; Table 1). Fescue Statistical analysis cover is greater in oviposition plots than in random plots, however, its structure coefficient of -0.343 ranks 9th out As multiple variables influence oviposition selection, we of the 22 variables in the analysis indicating that host plant were interested in quantifying where appropriate values of cover is not a primary driver of oviposition site selection the measured variables co-occurred, not just how mean (Table 1). The discriminant function classifies samples values compared across locations. To do this, we ran DFA better than random with a classification success of 80 %. with our oviposition and random plot data from 2009 using only the top six discriminating variables. This created a Oviposition habitat assessment method discriminant function that we then used to calculate the probability of oviposition occurring within each 2010 The average probability of oviposition in the two polygons meter-square plot based on all six of the sampled variables sampled in the high-density area of Scatter Creek is the (DISCRIM procedure, SAS 9.2). By using the average same (Table 2). The south polygon at Mima Mounds has a probability of oviposition in each half-hectare polygon as a probability of oviposition similar to the polygons at Scatter measure of habitat quality, we compared oviposition hab- Creek but the probability of oviposition in the north itat quality in the four polygons using a one-way ANOVA polygon of Mima Mounds is lower than that of the Scatter and post hoc Tukey tests (Minitab 15). Creek polygons (one-way ANOVA (a = 0.05): F = 3.52, df = 163, p = 0.017; Table 2). The discriminant function derived from the top discriminating variables from 2009 Results had an error rate of 16 % thereby classifying plots better than a random classification scheme. Oviposition surveys

During the 2009 flight season weather conditions were Discussion conducive for mardon skipper activity on 13 days during which we searched for females at 105 of the 217 WDFW Mardon skippers are host specialist butterflies in the Puget mardon skipper detection locations. We found females at prairies. Our observations of 86 of 88 eggs laid on 98 of the detection sites and witnessed oviposition at 88 Roemer’s fescue confirm prior speculation of fescue spe- locations. While patchy in distribution, oviposition loca- cialization in Puget populations. This is contrary to results tions are widespread thus confirming that butterflies are from Beyer and Schultz (2010) who found females to be ovipositing across the prairie (Fig. 2). All but two of the 88 generalists, laying eggs on 23 graminoid species across observed ovipositions were on Roemer’s fescue, a native nine meadows in the Washington Cascades. However, in perennial bunchgrass. The remaining two eggs were laid on meadows where Idaho fescue (Fescue idahoensis; closely 123 190 J Insect Conserv (2013) 17:183–194

Table 2 Classification analysis results Polygon Number of Average Tuft size #of % dead Vertical Moss Dist. To plots classified as probability of (cm3) fescue (category vegetation cover (%) nearest oviposition oviposition tufts 1–3) density (%) fescue tuft (cm)

Scatter Creek E 26 0.66a 8,889 ± 8,837 14 ± 5 2.0 ± 0.8 13 ± 14 58 ± 16 16 ± 8 Scatter Creek W 28 0.66a 8,908 ± 7,909 14 ± 7 1.9 ± 0.8 19 ± 20 52 ± 26 18 ± 11 Mima Mounds N 14 0.47b 13,131 ± 14,451 11 ± 5 2.4 ± 0.7 19 ± 21 62 ± 13 19 ± 12 Mima Mounds S 25 0.58ab 12,096 ± 12,663 11 ± 5 2.1 ± 0.8 6 ± 759± 24 19 ± 8 Results from discriminant function classification analysis of polygons sampled at Mima Mounds, WA and Scatter Creek, WA based. We sampled six habitat variables in 40 1-m2 plots in each of four ‘ ha (100 m 9 50 m) polygons. The number of plots per polygon classified as oviposition and the average probability of oviposition within each polygon are given. Locations with different superscripts have significantly different oviposition probabilities (ANOVA: df = 163, f = 3.52, p = 0.017). Variable values are mean ± 1SD related to Roemer’s fescue) was present, it was over- personal communication). We did not explicitly test this whelmingly preferred for oviposition indicating that some hypothesis; however, our findings of females selecting to montane populations exhibit oviposition specificity similar lay eggs in areas with low vertical vegetation density and to prairie populations. Local populations of other generalist little tall oatgrass cover (Table 1) are consistent with the butterfly species exhibit specialist tendencies (Singer 2004; hypothesis. The short, open vegetation structure of mardon Wiklund and Friberg 2008; Garcia-Barros and Fartmann oviposition habitat in the Puget prairies is similar to that of 2009). Localized host specialization can result from host other rare, temperate butterfly populations for which availability, host nutritional quality, parasitoid and or changes in vegetation structure have contributed to popu- predator communities supported by a particular host, lation declines (Thomas 1983; Thomas et al. 1986, 2009; ability of an herbivore to exploit a host’s chemical defen- Gutierrez et al. 1999;Mo¨llenbeck et al. 2009). ses, or host architecture which can influence microclimate Changes of both host plant architecture and local veg- and predator abundance (Lill et al. 2002; Reudler Talsma etation structure at oviposition sites can impact larval et al. 2008; Garcia-Barros and Fartmann 2009). survival in a number of ways including altering the For mardon skippers, simple presence of the hostplant, microclimate and predator community. Butterfly popula- Roemer’s fescue, is not adequate for oviposition. When tions are constrained by temperature (Crozier 2003, 2004) selecting oviposition locations, mardon skippers choose and are restricted to especially warm macro- and micro- small fescue tufts, at least half of the leaves of which are habitats in temperate climates (Thomas et al. 2001). The green. Selected tufts are surrounded by moss, and neigh- small tufts and open habitat structure selected by ovipos- boring tall oatgrass cover and density of vertical vegetation iting females are likely to correspond to the warmest are low. The sparse, open vegetation structure of oviposi- locations in the prairie (Forsberg 1987; Stoutjesdijk and tion locations is similar to that of historic glacial outwash Barkman 1992). Weather in the Puget lowlands is cool and prairies (Chappell and Crawford 1997). Invasion of the cloudy with intermittent sun breaks for much of the year, Puget prairies has dramatically altered their shortgrass particularly in spring when overwintering larvae complete structure as both scotch broom and tall oatgrass can reach development and pupate (February–April average max heights of over 2 m. During oviposition surveys we often temp: 12.1 °C, average min temp: 1.2 °C; Western lost females when they encountered large swaths of tall Regional Climate Center 2010). In these conditions, loca- oatgrass primarily due to behavioral changes—mainly tions where ectothermic larvae can take advantage of solar increased velocity. We did, however, follow one female in heating may result in increased foraging and faster devel- fast, straight, flight across 70 m of tall oatgrass to a mossy, opment time, potentially increasing individual survival open fescue patch where she promptly laid an egg. (Bonebrake et al. 2010). Although anecdotal, this observation suggests that even Another explanation for selection of oviposition loca- when fescue is hiding below a canopy of tall oatgrass tions is predator avoidance. Larger, more structurally females avoid ovipositing in heavily invaded areas, as has complex grass tufts support a greater density of inverte- been observed in other butterfly species (Severns 2008). brate predators (Reid and Hochuli 2007) that may play an WDFW biologists hypothesize that changes from the open important role in early instar larval predation (Wiklund and vegetation structure of historic plant communities to the Friberg 2008). Additionally, larval predation rates are complex structure of invaded areas has led to mardon influenced by vegetation density; with predation being skipper population declines in Puget prairies (A. Potter, lowest in isolated host plants and greatest where there is a

123 J Insect Conserv (2013) 17:183–194 191 high density of vegetation surrounding the host plant historic vegetation structure and is unlikely to be imple- (Wiklund and Friberg 2008). Mardon skipper larvae mented in prairie restoration. Instead, common butterfly developing in large fescue tufts or tufts surrounded by habitat management techniques in the region include grass vegetation may be more likely to encounter predators than specific and broad-spectrum herbicide application, mow- those in small, isolated tufts. ing, and prescribed burning (Schultz et al. 2011). Of these The selection of oviposition sites with 56 % moss cover techniques only fire has the ability to reduce the basal area reinforces the need for open space at oviposition locations. of individual fescue plants (Conrad and Poulton 1966; Bare ground and lichen crusts occupied the interstitial areas Tveten and Fonda 1999). Prescribed burning is commonly of historic prairies (Chappell and Kagan 2001). In modern used in grassland restoration projects with goals of creating prairie environments, bare ground is quickly invaded, most habitat for rare species (Warren 1991; Schultz and Crone often by non-native species. The mat-like carpet of moss 1998;Mo¨llenbeck et al. 2009). One challenge of using fire (primarily Racomitrium canescens) that has developed in to improve habitat for endangered butterflies is that larvae the absence of fire inhibits plant germination and estab- rarely survive burning (Swengel 2001). However, when lishment (Morgan 2006; Stanley et al. 2011), effectively patchy burning strategies are employed that allow re-col- maintaining weed-free open spaces. We observed mardon onization of burned areas by adults, the benefits of pre- skippers actively using the moss spaces between fescue scribed burning may potentially outweigh the costs (Dana tufts for basking. Males used the moss space when perch- 1991; Schultz and Crone 1998; Relf and New 2009). A ing and chasing potential mates, and searched for females well-planned burn relies on understanding the life history, in these open mossy spots when patrolling (E. Henry per- habitat requirements, and distribution within a site of sonal observation). These observations suggest that adults species that will be impacted (Dana 1991; Schultz and use moss for thermoregulation and mate finding much in Crone 1998). Without sufficient species specific informa- the way other species use bare ground and open areas tion, burning, or any intensive management, can be cata- (Clench 1966; Wickman 2009). strophic for small populations (Warren 1991; Konvicka Maintaining the vegetation structure characteristic of et al. 2008). mardon skipper oviposition habitat in Puget prairies requires continued invasive species management. Although scotch broom cover was not one of the primary discrimi- Conclusion nating variables it is still significantly lower at oviposition locations than at random locations and its management is Oviposition habitat requirements of mardon skippers in the not unimportant. Intense, ongoing scotch broom control at Puget prairies are distinct from those in the southern Scatter Creek has reduced and prevented the expansion of Washington Cascades. In the Cascades, female mardon scotch broom across the site, especially in areas occupied skippers oviposit on larger (greater cover) graminoids by the butterflies. Without such control, it is likely that where there is more vegetation cover and less bare ground scotch broom cover would have a greater influence on (Beyer and Schultz 2010). These divergent results, in part, oviposition site selection because of the dramatic structural reflect differences in climate between the two regions. In changes associated with its invasion. the mountains, larvae are buried by snow from November Removal of invasive species alone, however, will not be to June. When not under the snow, mountain larvae are sufficient when restoring mardon skipper habitat. Despite exposed to greater daily temperature swings and perhaps the near absence of tall oatgrass and scotch broom, our greater graminoid cover buffers these extreme tempera- habitat assessment analysis showed that the north polygon tures, keeping caterpillars from both freezing and over- at Mima Mounds does not have the same probability of heating (Stoutjesdijk and Barkman 1992). Additionally, oviposition as high quality areas at Scatter Creek. Fescue mountain larvae may be exposed to higher levels of solar tufts in both polygons at Mima Mounds are on average radiation than prairie larvae thus causing females to target 50 % larger than those in the high-density area of Scatter different host structure for oviposition in the two popula- Creek, and twice as big as selected oviposition tufts tions (Anthes et al. 2008). (Tables 1, 2). This pattern illustrates the potential need for Given the contrasting structural habitat requirements tuft size management in areas of prairie where tall oatgrass between the two regions, application of results from Beyer and scotch broom are presently under control. and Schultz’s (2010) work in the mountains to mardon Maintenance of small host plants and short sward skipper habitat restoration projects in the Puget prairies heights for butterflies in Europe has primarily been could result in the enhancement of structural attributes achieved through grazing (WallisDeVries and Raemakers away from those that promote mardon skipper oviposition 2001; Eichel and Fartmann 2008; Thomas et al. 2009). In in these prairies. Results from both of these studies Puget prairies, grazing was not instrumental in maintaining emphasize the importance of understanding local behavior 123 192 J Insect Conserv (2013) 17:183–194 and habitat use of rare species populations where restora- field season with little time off and collected fabulous data. Special tion activities occur (Dennis 2004, 2010). thanks to Ann Potter, Dave Hays, Gail Olson, and Mary Linders of WDFW and David Wilderman of WA Departement of Natural Butterfly habitat quality at a site is often evaluated by Resources for sharing their Puget prairie and mardon skipper wisdom hostplant abundance alone (Lipman et al. 1999; Fartmann and for being eager collaborators. Additionally we thank two anon- 2006; Relf and New 2009; Bartel et al. 2010) even though ymous reviewers for their comments. This work was funded by the all potential hosts in a site may not be suitable for ovipo- U. S. Army Compatible Use Buffer Program at Joint Base Lewis McChord, Washington State University Vancouver, and the following sition and larval development (Thomas 1984; Eichel and grants and fellowships awarded to Erica H. Henry: National Science Fartmann 2008; Bonebrake et al. 2010; Dennis 2010). Foundation GK-12 fellowship (NSF Award number 07-42561 to Results of our habitat assessment analysis confirm that our G. Rollwagen-Bollens), The Xerces Society’s DeWind Award for method works to pick up important differences in ovipo- Lepidoptera Conservation, a Prairie Biotic Research grant for prairie and savannah research, and a WSU Robert Lane Fellowship in sition habitat quality in multiple locations, and can be Environmental Studies. useful to land managers when selecting target areas for habitat management and measuring effects of those activ- ities. It is important to note that another aspect of habitat References suitability not accounted for herein is the area of suitable habitat (patch size) necessary for mardon skipper persis- Anthes N, Fartmann T, Hermann G (2008) The Duke of Burgundy tence. In a large prairie site such as Mima mounds it is butterfly and its dukedom: larval niche variation in Hamearis unlikely that the entire site must contain suitable mardon lucina across Central Europe. J Insect Conserv 12:3–14 skipper oviposition habitat, but rather that a large enough Barinaga M (1990) Where have all the froggies gone? Science patch, or series of patches, within the site does. Prior to 247:1033–1034 Bartel RA, Haddad NM, Wright JP (2010) Ecosystem engineers reintroduction this must be assessed. Currently the data maintain a rare species of butterfly and increase plant diversity. needed to determine appropriate patch size and distribu- Oikos 119:883–890 tion—such as dispersal behavior and demography (Schultz Beyer L, Black SH (2006) Site utilization by adults and larvae of and Crone 2005)—are not available; this would be a logical mardon skipper butterfly (Polites mardon) at four sites in Washington and Oregon. Report to the Forest Service and next-step toward the conservation of the species. Bureau of Land Management from the Xerces Society, 73 pp The need for restoration and conservation of specific Beyer LJ, Schultz CB (2010) Oviposition selection by a rare grass habitat attributes is not unique to rare butterflies. Micro- skipper Polites mardon in montane habitats: advancing ecolog- habitat attributes such as vegetation structure and micro- ical understanding to develop conservation strategies. Biol Conserv 143:862–872 climate are also critical to the survival of rare insects such Boggs CL, Ross CL (1993) The effect of adult food limitation on life as a flightless bush-cricket (Methoptera brachyptera), a history traits in Speyeria mormonia (Lepidoptera: Nymphalidae). long-horned beetle (Rosalia alpina), and a dung beetle Ecology 74:433–441 (Onthophagus gibbulus) (Roslin et al. 2009; Poniatowski Bonebrake TC, Boggs CL, McNally JM, Ranganathan J, Ehrlich PR (2010) Oviposition behavior and offspring performance in and Fartmann 2010; Russo et al. 2010). The research herbivorous insects: consequences of climatic and habitat needed to elucidate these fine-scale habitat requirements is heterogeneity. Oikos 119:927–934 often seen as time consuming and expensive; however, Chappell CB, Crawford RC (1997) Native vegetation of the south truly understanding the ecology of a rare species is fun- Puget Sound prairie landscape. In: Dunn PV, Ewing K (eds) Ecology and conservation of the South Puget Sound prairie damental to successful conservation (Miller and Hobbs landscape. The Nature Conservancy, Seattle, pp 107–122 2007; Schultz and Crone 2008; Morrison 2009). Without Chappell CB, Kagan J (2001) Westside grasslands. In: Johnson DH, the kind of detailed information gleaned by scrutinizing O’Neil TA (eds) Wildlife-Habitat relationships in Oregon and individuals’ behaviors and habitat, habitat conservation and Washington. Oregon State University Press, Corvallis, pp 41–43 Clench HK (1966) Behavioral thermoregulation in butterflies. Ecol- restoration projects run the risk of missing the mark, or ogy 47:1021–1034 wasting resources on activities that turn out to be detri- Conrad CE, Poulton CE (1966) Effect of a wildfire on Idaho fescue mental to the target species (Dennis 2010). Our approach of and bluebunch wheatgrass. J Range Manage 19:138–141 translating individual habitat use, to specific structural Crawford RC, Hall H (1997) Changes in the south Puget prairie landscape. In: Dunn PV, Ewing K (eds) Ecology and conserva- habitat requirements, to a habitat evaluation technique tion of the South Puget Sound prairie landscape. The Nature advances the understanding of an understudied butterfly Conservancy, Seattle, pp 11–16 family, and greatly increases our ability to target habitat Crone EE, Schultz CB (2003) Movement behavior and minimum restoration where it is most needed and useful for the patch size for butterfly population persistence. In: Boggs CL, Watt WB, Ehrlich PR (eds) Butterflies: ecology and evolution persistence of the species. taking flight. The University of Chicago Press, Chicago, pp 560–576 Acknowledgments We thank Scott Pearson and John Bishop for Crozier L (2003) Winter warming facilitates range expansion: cold their feedback on experimental design and previous drafts. Justin tolerance of the butterfly Atalopedes campestris. Oecologia Kirsch and Brad Gill worked numerous long, hot days throughout the 135:648–656

123 J Insect Conserv (2013) 17:183–194 193

Crozier L (2004) Warmer winters drive butterfly range expansion by Jepsen S, Lauvray L, Black SH (2008) Xerces society surveys for increasing survivorship. Ecology 85:231–241 Polites mardon mardon in the Naches Ranger District Dana RP (1991) Conservation management of the prairie skippers (Wenatchee National Forest) of Washington. Report to the U. Hesperia dacotae and Hesperia ottoe: basic biology and threat of S. Forest Service, 15 pp mortality during prescribed burning in spring. Minnesota Agri- Karlsson B, Wiklund C (2005) Butterfly life history and temperature cultural Experiment Station Bulletin 594, 63 pp adaptations: dry open habitats select for increased fecundity and Dennehy C, Alverson ER, Anderson HE, Clements DR, Gilbert R, longevity. J Anim Ecol 74:99–104 Kaye TN (2011) Management strategies for invasive plants in Konvicka M, Benes J, Cizek O, Kopecek F, Konvicka O, Vitaz L Pacific Northwest Prairies, savannas, and oak woodlands. (2008) How too much care kills species: Grassland reserves, Northwest Sci 85:329–351 agri-environmental schemes and extinction of Colias myrmidone Dennis RLH (2004) Butterfly habitats, broad-scale biotope affilia- (Lepidoptera: Pieridae) from its former stronghold. J Insect tions, and structural exploitation of vegetation at finer scales: the Conserv 12:519–525 matrix revisited. Ecol Entomol 29:744–752 Lill JT, Marquis RJ, Ricklefs RE (2002) Host plants influence Dennis RLH (2010) A resource-based habitat view for conservation. parasitism of forest caterpillars. Nature 417:170–173 Butterflies in the British landscape. Wiley-Blackwell, West Sussex Lipman A, Longcore T, Mattoni R, Zhang Y (1999) Habitat Dennis RLH, Shreeve TG, Van Dyck H (2003) Towards a functional evaluation and reintroduction planning for the endangered Palos resource-based concept for habitat: a butterfly biology view- Verdes blue butterfly. Final Technical Report to California point. Oikos 102:417–426 Department of Fish and Game, San Diego, 47 pp Dennis RLH, Shreeve TG, Van Dyck H (2006) Habitats and Longcore T, Lam CS, Kobernus P, Polk E, Wilson JP (2010) resources: the need for a resource-based definition to conserve Extracting useful data from imperfect monitoring schemes: butterflies. Biodivers Conserv 15:1943–1966 Endangered butterflies at San Bruno Mountain, San Mateo Dennis RLH, Hardy PB, Shreeve TG (2008) The importance of County, California (1982–2000) and implications for habitat resource databanks for conserving insects: a butterfly biology management. J Insect Conserv 14:335–346 perspective. J Insect Conserv 12:711–719 Mattoon SO, Emmel JF, Emmel TC (1998) The distribution of Polites Eichel S, Fartmann T (2008) Management of calcareous grasslands mardon (Lepidoptera: Hesperiidae) in North America, and for Nickerl’s fritillary (Melitaea aurelia) has to consider habitat description of a new subspecies from southern Oregon. In: requirements of the immature stages, isolation, and patch area. Emmel TC (ed) Systematics of Western North American J Insect Conserv 12:677–688 butterflies. Mariposa Press, Gainsville, pp 767–774 Fartmann T (2006) Oviposition preferences, adjacency of old McGarigal K, Cushman S, Stafford S (2000) Multivariate statistics for woodland and isolation explain the distribution of the Duke of wildlife ecology and research. Springer Science ? Business Burgandy butterfly (Hamearis lucina) in calcareous grasslands in Media, Inc., New York central Germany. Ann Zool Fenn 43:335–347 Miller JR, Hobbs RJ (2007) Habitat restoration—do we know what Forsberg J (1987) Size discrimination among conspecific hostplants in we’re doing? Restor Ecol 15:382–390 two Pierid butterflies: Pieris napi L. and Pontia daplidice L. Mo¨llenbeck V, Hermann G, Fartmann T (2009) Does prescribed Oecologia 72:52–57 burning mean a threat to the rare satyrine butterfly Hipparchia Garcia-Barros E, Fartmann T (2009) Butterfly oviposition: sites, fagi? Larval-habitat preferences give the answer. J Insect behaviour and modes. In: Settele J, Shreeve TG, Konvicka M, Conserv 13:77–87 Van Dyck H (eds) Ecology of butterflies in Europe. Cambridge Morgan JW (2006) Bryophyte mats inhibit germination of non-native University Press, Cambridge, pp 29–42 species in burnt temperate native grassland remnants. Biol Goodall DW (1952) Some considerations in the use of point quadrates Invasions 8:159–168 for the analysis of vegetation. Aust J Sci Res Ser B 5:1–41 Morrison ML (2009) Wildlife restoration: synthesis. In: Morrison ML Griffith B, Scott JM, Carpenter JW, Reed C (1989) Translocation as a (ed) Restoring wildlife ecological concepts and practical appli- species conservation tool: status and strategy. Science 245: cations. Island Press, Washington, DC, pp 287–300 477–480 New TR, Pyle RM, Thomas JA, Thomas CD, Hammond PC (1995) Gutierrez D, Thomas CD, Leon-Cortes JL (1999) Dispersal, distri- Butterfly conservation management. Annu Rev Entomol 40: bution, patch network and metapopulation dynamics of the dingy 57–83 skipper butterfly (Erynnis tages). Oecologia 121:506–517 Newmark WD (1996) Insularization of Tanzanian parks and the local Hanski I (1991) Single-species metapopulation dynamics. In: Gilpin extinction of large mammals. Conserv Biol 10:1549–1556 M, Hanski I (eds) Metapopulation dynamics: empirical and Noss RF, LaRoe ET III, Scott JM (1995) Endangered ecosystems of theoretical investigations. Academic Press, London, pp 17–38 the United States: a preliminary assessment of loss and Hanski I (2003) Biology of extinctions in butterfly metapopulations. degradation. United States Geological Survey: Biological In: Boggs CL, Watt WB, Ehrlich PR (eds) Butterflies: ecology Resources and evolution taking flight. The University of Chicago Press, Nylin S, Gotthard K (1998) Plasticity in life-history traits. Annu Rev Chicago, pp 577–602 Entomol 43:63–83 Hays D, Potter A, Thompson C, Dunn P (2000) Critical habitat Olson G (2010) ACUB Prairie assessment progress report. For components for four rare south Puget Sound grassland butter- ACUB/Prairie Legacy Partners. Washington Department of Fish flies. Final Report to Washington Department of Fish and and Wildlife. Olympia Wildlife. Olympia, 39 pp Panzer R, Schwartz MW (1998) Effectiveness of a vegetation-based Henry EH (2010) A first step towards successful habitat restoration approach to insect conservation. Conserv Biol 12:693–702 and reintroduction: understanding oviposition site selection of an Poniatowski D, Fartmann T (2010) What determines the distribution imperiled butterfly, mardon skipper. Master’s thesis. Washington of a flightless bush-cricket (Merioptera brachyptera)ina State University, Vancouver fragmented landscape? J Insect Conserv 14:637–645 IUCN (1998) IUCN guidelines for re-introductions. Prepared by the Potter AE, Olson G (2009) Monitoring mardon skipper at scatter IUCN/SSC re-introduction specialist group. IUCN, Gland, creek wildlife area. Army compatible use buffer (ACUB) project Switzerland, 10 pp proposal. Washington Department of Fish and Wildlife, Olympia

123 194 J Insect Conserv (2013) 17:183–194

Potter AE, Fleckenstein J, Richardson S, Hays D (1999) Washington Stoutjesdijk P, Barkman JJ (1992) Microclimate, vegetation and State status report for the mardon skipper. Washington Depart- fauna. Opulus Press AB, Knivsta ment of Fish and Wildlife, Olympia Swengel AB (2001) A literature review of insect responses to fire, Quinn GP, Keough MJ (2002) Experimental design and data analysis compared to other conservation managements of open habitat. for biologists. Cambridge University Press, Cambridge Biodivers Conserv 10:1141–1169 Reid AM, Hochuli DF (2007) Grassland invertebrate assemblages in Sykes JM, Horrill AD, Mountford MD (1983) Use of visual cover managed landscapes: effect of host plant and microhabitat assessments as quantitative estimators of some British woodland architecture. Aust Ecol 32:708–718 taxa. J Ecol 71:437–450 Relf MC, New TR (2009) Conservation needs of the Altona skipper Thomas JA (1983) The ecology and conservation of Lysandra butterfly, Hesperilla flavescens flavescens Waterhouse (Lepi- bellargus (Lepidoptera: Lycaenidae) in Britain. J Appl Ecol doptera: Herperiidae), near Melbourne, Victoria. J Insect Con- 20:59–83 serv 13:143–149 Thomas JA (1984) The conservation of butterflies in temperate Reudler Talsma JH, Biere A, Harvey JA, van Nouhuys S (2008) countries: past efforts and lessons for the future. In: Vane-Wright Oviposition cues for a specialist butterfly-plant chemistry and RI, Ackery PR (eds) The biology of butterflies. Academic Press, size. J Chem Ecol 34:1202–1212 London, pp 333–353 Roslin T, Avomaa T, Leonard ML, Ovaskainen O (2009) Some like it Thomas J, Thomas C, Simcox D, Clarke R (1986) Ecology and hot: microclimatic variation affects the abundance and move- declining status of the silver-spotted skipper butterfly (Hesperia ments of a critically endangered dung beetle. Insect Conserv comma) in Britain. J Appl Ecol 23:365–380 Diver 2:232–241 Thomas CD, Bodsworth EJ, Wilson RJ, Simmons MT, Davies ZG, Russo D, Cistrone L, Garonna AP (2010) Habitat selection by the Musche M, Conradt L (2001) Ecological and evolutionary highly endangered long-horned beetle Rosalia alpina in Southern processes at expanding range margins. Nature 411:577–581 Europe: a multiple spatial scale assessment. J Insect Conserv Thomas JA, Simcox DJ, Clarke RT (2009) Successful conservation of 15:685–693 a threatened Maculinea butterfly. Science 325:80–83 Scheffe H (1959) The analysis of variance. Wiley, New York Thomas JA, Simcox DJ, Hovestadt T (2011) Evidence based Schultz CB, Crone EE (1998) Burning prairie to restore butterfly conservation of butterflies. J Insect Conserv 15:241–258 habitat: a modeling approach to management tradeoffs for the Tveten RK, Fonda RW (1999) Fire effects on prairies and oak Fender’s blue. Restor Ecol 6:244–252 woodlands on Fort Lewis, Washington. Northwest Sci 73: Schultz CB, Crone EE (2005) Patch size and connectivity thresholds 145–158 for butterfly habitat restoration. Conserv Biol 19:887–896 USFWS (2011) Endangered and threatened wildlife and plants; Schultz CB, Crone EE (2008) Using ecological theory to advance review of native species that are candidates for listing as butterfly conservation. Isr J Ecol Evol 54:63–68 endangered or threatened; annual notice of findings on resub- Schultz CB, Henry E, Carleton A, Hicks T, Thomas R, Potter A, mitted petitions; annual description of progress on listing Collins M, Linders M, Fimbel C, Black SH, Anderson H, Diehl actions. Fed Reg 76:66370–66439 G, Hamman S, Gilbert R, Foster J, Hays D, Page N, Heron J, WallisDeVries MF, Raemakers I (2001) Does extensive grazing Kroeker N, Webb C, Reader B (2011) Conservation of prairie- benefit butterflies in coastal dunes? Restor Ecol 9:179–188 oak butterflies in Oregon, Washington, and British Columbia. Warren MS (1991) The successful conservation of an endangered Northwest Sci 85:361–388 species, the heath fritillary butterfly Mellicta athalia, in Britain. Settele J, Kuhn E (2009) Insect conservation. Science 325:41–42 Biol Conserv 55:37–56 Severns PM (2008) Exotic grass invasion impacts fitness of an Wenzel M, Schmitt T, Weitzel M, Seitz A (2006) The severe decline endangered prairie butterfly, Icaricia icarioides fenderi. J Insect of butterflies on western German calcareous grasslands during Conserv 12:651–661 the last 30 years: a conservation problem. Biol Conserv 128: Shreeve TG, Dennis RLH, Van Dyck H (2004) Resources, habitats 542–552 and metapopulations—whither reality? Oikos 106:404–408 Western Regional Climate Center (2010) Western U.S. climate Singer M (2004) Measurement, correlates, and importance of historical summaries, Washington, Olympia Airport Station, oviposition preference in the life of checkerspots. In: Ehrlich 1948–2009. URL http://www.wrcc.dri.edu/cgi-bin/cliMAIN.pl? PR, Hanski I (eds) On the wings of checkerspots: a model system wa6114. Accessed on 10 Oct 2010 for population biology. Oxford University Press, New York, Wickman P (2009) Thermoregulation and habitat use in butterflies. pp 112–137 In: Settele J, Shreeve TG, Konvicka M, VanDyck H (eds) Stanley AG, Kaye TN, Dunwiddie PW (2011) Multiple treatment Ecology of butterflies in Europe. Cambridge University Press, combinations and seed addition increase abundance and diversity Cambridge, pp 55–61 of native plants in Pacific Northwest prairies. Ecol Restor Wiklund C, Friberg M (2008) Enemy-free space and habitat-specific 29:35–44 host specialization in a butterfly. Oecologia 157:287–294 Storm L, Shebitz D (2006) Evaluating the purpose, extent, and Williams B (1983) Some observations of the use of discriminant ecological restoration applications of indigenous burning prac- analysis in ecology. Ecology 64:1283–1291 tices in southwestern Washington. Ecol Restor 24:256–258

123